[ltp] HPET timer on thinkpads?

Tue, 23 Oct 2007 08:10:19 -0700

This is a multi-part message in MIME format.
--------------050003070703000100010709
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit

I found the no_hz config parm prevented the cpu from running at lower 
clock speeds.

Clark Williams wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Henrique de Moraes Holschuh wrote:
>   
>> On Tue, 23 Oct 2007, Igor V. Rafienko wrote:
>>     
>>> on Oct 23, 2007, 07:37, Nils wrote:
>>>       
>>>>> that draws about 1.3W *more* than 2.6.22 without the hpet-patches. Does 
>>>>> anyone else experience that?
>>>>>           
>>>> Yes, I've the same problem with the T41p and 2.6.23.1-hrt3. (I thought it 
>>>> was a problem with slub or slab allocator, but the power consumption is 
>>>> the same with both allocators.)
>>>>         
>>> Hmm... this cannot possibly be the intention of the patch.
>>>       
>> Test pure 2.6.23.1 first.  Or don't, 2.6.23 has some bad problems in ACPI,
>> that are probably going to get fixed in 2.6.23.2.
>>
>> But -hrt is NOT about power consumption control, it is about real-time
>> behaviour.
>>
>>     
>
> Not strictly true. The -hrt patch is mostly about improving timer granularity above
> the standard system clock tick. In addition the NO_HZ mechanism really is targeted at
> reducing power consumption, by reducing the number of interrupts the system has to
> service when in idle state.
>
> The -rt patch (which includes the -hrt patch) is about real-time behavior.
>
> Clark
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.7 (GNU/Linux)
> Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org
>
> iD8DBQFHHgOHHyuj/+TTEp0RAuv5AJ43+bGbxtBzu7rpTK8HsavJdNaSpQCfQnB3
> Y36EQZW6Kx0DLMMQDKenDeY=
> =7+Zz
> -----END PGP SIGNATURE-----
>   

--------------050003070703000100010709
Content-Type: text/html; charset=UTF-8
Content-Transfer-Encoding: 7bit

<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html>
<head>
  <meta content="text/html;charset=UTF-8" http-equiv="Content-Type">
  <title></title>
</head>
<body bgcolor="#ffffff" text="#000000">
I found the no_hz config parm prevented the cpu from running at lower
clock speeds. <br>
<br>
Clark Williams wrote:
<blockquote cite="mid:471E0387.1020701@redhat.com" type="cite">
  <pre wrap="">-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Henrique de Moraes Holschuh wrote:
  </pre>
  <blockquote type="cite">
    <pre wrap="">On Tue, 23 Oct 2007, Igor V. Rafienko wrote:
    </pre>
    <blockquote type="cite">
      <pre wrap="">on Oct 23, 2007, 07:37, Nils wrote:
      </pre>
      <blockquote type="cite">
        <blockquote type="cite">
          <pre wrap="">that draws about 1.3W *more* than 2.6.22 without the hpet-patches. Does 
anyone else experience that?
          </pre>
        </blockquote>
        <pre wrap="">Yes, I've the same problem with the T41p and 2.6.23.1-hrt3. (I thought it 
was a problem with slub or slab allocator, but the power consumption is 
the same with both allocators.)
        </pre>
      </blockquote>
      <pre wrap="">Hmm... this cannot possibly be the intention of the patch.
      </pre>
    </blockquote>
    <pre wrap="">Test pure 2.6.23.1 first.  Or don't, 2.6.23 has some bad problems in ACPI,
that are probably going to get fixed in 2.6.23.2.

But -hrt is NOT about power consumption control, it is about real-time
behaviour.

    </pre>
  </blockquote>
  <pre wrap=""><!---->
Not strictly true. The -hrt patch is mostly about improving timer granularity above
the standard system clock tick. In addition the NO_HZ mechanism really is targeted at
reducing power consumption, by reducing the number of interrupts the system has to
service when in idle state.

The -rt patch (which includes the -hrt patch) is about real-time behavior.

Clark
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.7 (GNU/Linux)
Comment: Using GnuPG with Fedora - <a class="moz-txt-link-freetext" href="http://enigmail.mozdev.org">http://enigmail.mozdev.org</a>

iD8DBQFHHgOHHyuj/+TTEp0RAuv5AJ43+bGbxtBzu7rpTK8HsavJdNaSpQCfQnB3
Y36EQZW6Kx0DLMMQDKenDeY=
=7+Zz
-----END PGP SIGNATURE-----
  </pre>
</blockquote>
</body>
</html>

--------------050003070703000100010709--